1 use crate::post::SiteMetadata;
2 use encoding::{all::encodings, DecoderTrap};
3 use lemmy_db_schema::newtypes::DbUrl;
6 settings::structs::Settings,
10 use percent_encoding::{utf8_percent_encode, NON_ALPHANUMERIC};
11 use reqwest_middleware::ClientWithMiddleware;
12 use serde::Deserialize;
17 /// Fetches the post link html tags (like title, description, image, etc)
18 #[tracing::instrument(skip_all)]
19 pub async fn fetch_site_metadata(
20 client: &ClientWithMiddleware,
22 ) -> Result<SiteMetadata, LemmyError> {
23 info!("Fetching site metadata for url: {}", url);
24 let response = client.get(url.as_str()).send().await?;
26 // Can't use .text() here, because it only checks the content header, not the actual bytes
27 // https://github.com/LemmyNet/lemmy/issues/1964
28 let html_bytes = response.bytes().await.map_err(LemmyError::from)?.to_vec();
30 let tags = html_to_site_metadata(&html_bytes)?;
35 fn html_to_site_metadata(html_bytes: &[u8]) -> Result<SiteMetadata, LemmyError> {
36 let html = String::from_utf8_lossy(html_bytes);
38 // Make sure the first line is doctype html
44 .ok_or_else(|| LemmyError::from_message("No lines in html"))?
47 if !first_line.starts_with("<!doctype html>") {
48 return Err(LemmyError::from_message(
49 "Site metadata page fetch is not DOCTYPE html",
53 let mut page = HTML::from_string(html.to_string(), None)?;
55 // If the web page specifies that it isn't actually UTF-8, re-decode the received bytes with the
56 // proper encoding. If the specified encoding cannot be found, fall back to the original UTF-8
58 if let Some(charset) = page.meta.get("charset") {
59 if charset.to_lowercase() != "utf-8" {
60 if let Some(encoding_ref) = encodings().iter().find(|e| e.name() == charset) {
61 if let Ok(html_with_encoding) = encoding_ref.decode(html_bytes, DecoderTrap::Replace) {
62 page = HTML::from_string(html_with_encoding, None)?;
68 let page_title = page.title;
69 let page_description = page.description;
71 let og_description = page
75 .map(std::string::ToString::to_string);
80 .map(std::string::ToString::to_string);
85 .and_then(|ogo| Url::parse(&ogo.url).ok());
86 let og_embed_url = page
90 .and_then(|v| Url::parse(&v.url).ok());
93 title: og_title.or(page_title),
94 description: og_description.or(page_description),
95 image: og_image.map(Into::into),
96 embed_video_url: og_embed_url.map(Into::into),
100 #[derive(Deserialize, Debug, Clone)]
101 pub(crate) struct PictrsResponse {
102 files: Vec<PictrsFile>,
106 #[derive(Deserialize, Debug, Clone)]
107 pub(crate) struct PictrsFile {
110 delete_token: String,
113 #[derive(Deserialize, Debug, Clone)]
114 pub(crate) struct PictrsPurgeResponse {
118 #[tracing::instrument(skip_all)]
119 pub(crate) async fn fetch_pictrs(
120 client: &ClientWithMiddleware,
123 ) -> Result<PictrsResponse, LemmyError> {
124 let pictrs_config = settings.pictrs_config()?;
125 is_image_content_type(client, image_url).await?;
127 let fetch_url = format!(
128 "{}image/download?url={}",
130 utf8_percent_encode(image_url.as_str(), NON_ALPHANUMERIC) // TODO this might not be needed
133 let response = client
135 .timeout(REQWEST_TIMEOUT)
139 let response: PictrsResponse = response.json().await.map_err(LemmyError::from)?;
141 if response.msg == "ok" {
144 Err(LemmyError::from_message(&response.msg))
148 /// Purges an image from pictrs
149 /// Note: This should often be coerced from a Result to .ok() in order to fail softly, because:
150 /// - It might fail due to image being not local
151 /// - It might not be an image
152 /// - Pictrs might not be set up
153 pub async fn purge_image_from_pictrs(
154 client: &ClientWithMiddleware,
157 ) -> Result<(), LemmyError> {
158 let pictrs_config = settings.pictrs_config()?;
159 is_image_content_type(client, image_url).await?;
161 let alias = image_url
163 .ok_or_else(|| LemmyError::from_message("Image URL missing path segments"))?
165 .ok_or_else(|| LemmyError::from_message("Image URL missing last path segment"))?;
167 let purge_url = format!("{}/internal/purge?alias={}", pictrs_config.url, alias);
169 let pictrs_api_key = pictrs_config
171 .ok_or_else(|| LemmyError::from_message("pictrs_api_key_not_provided"))?;
172 let response = client
174 .timeout(REQWEST_TIMEOUT)
175 .header("x-api-token", pictrs_api_key)
179 let response: PictrsPurgeResponse = response.json().await.map_err(LemmyError::from)?;
181 if response.msg == "ok" {
184 Err(LemmyError::from_message(&response.msg))
188 /// Both are options, since the URL might be either an html page, or an image
189 /// Returns the SiteMetadata, and a Pictrs URL, if there is a picture associated
190 #[tracing::instrument(skip_all)]
191 pub async fn fetch_site_data(
192 client: &ClientWithMiddleware,
195 ) -> (Option<SiteMetadata>, Option<DbUrl>) {
199 // Ignore errors, since it may be an image, or not have the data.
200 // Warning, this may ignore SSL errors
201 let metadata_option = fetch_site_metadata(client, url).await.ok();
203 // Fetch pictrs thumbnail
204 let pictrs_hash = match &metadata_option {
205 Some(metadata_res) => match &metadata_res.image {
206 // Metadata, with image
207 // Try to generate a small thumbnail if there's a full sized one from post-links
208 Some(metadata_image) => fetch_pictrs(client, settings, metadata_image)
210 .map(|r| r.files[0].file.clone()),
211 // Metadata, but no image
212 None => fetch_pictrs(client, settings, url)
214 .map(|r| r.files[0].file.clone()),
216 // No metadata, try to fetch the URL as an image
217 None => fetch_pictrs(client, settings, url)
219 .map(|r| r.files[0].file.clone()),
222 // The full urls are necessary for federation
223 let pictrs_thumbnail = pictrs_hash
226 "{}/pictrs/image/{}",
227 settings.get_protocol_and_hostname(),
235 (metadata_option, pictrs_thumbnail.map(Into::into))
237 None => (None, None),
241 #[tracing::instrument(skip_all)]
242 async fn is_image_content_type(client: &ClientWithMiddleware, url: &Url) -> Result<(), LemmyError> {
243 let response = client.get(url.as_str()).send().await?;
247 .ok_or_else(|| LemmyError::from_message("No Content-Type header"))?
249 .starts_with("image/")
253 Err(LemmyError::from_message("Not an image type."))
257 pub fn build_user_agent(settings: &Settings) -> String {
261 settings.get_protocol_and_hostname()
267 use crate::request::{build_user_agent, fetch_site_metadata, SiteMetadata};
268 use lemmy_utils::settings::SETTINGS;
271 // These helped with testing
273 async fn test_site_metadata() {
274 let settings = &SETTINGS.clone();
275 let client = reqwest::Client::builder()
276 .user_agent(build_user_agent(settings))
280 let sample_url = Url::parse("https://gitlab.com/IzzyOnDroid/repo/-/wikis/FAQ").unwrap();
281 let sample_res = fetch_site_metadata(&client, &sample_url).await.unwrap();
284 title: Some("FAQ · Wiki · IzzyOnDroid / repo · GitLab".to_string()),
286 "The F-Droid compatible repo at https://apt.izzysoft.de/fdroid/".to_string()
289 Url::parse("https://gitlab.com/uploads/-/system/project/avatar/4877469/iod_logo.png")
293 embed_video_url: None,
300 // fn test_pictshare() {
301 // let res = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpg");
302 // assert!(res.is_ok());
303 // let res_other = fetch_pictshare("https://upload.wikimedia.org/wikipedia/en/2/27/The_Mandalorian_logo.jpgaoeu");
304 // assert!(res_other.is_err());